MOX Colloquia

Abstract

Discovering fundamental laws governing systems from observational data has long been a hallmark of scientific inquiry. In this talk, I will discuss how recent advances in AI and machine learning enable the automated discovery of scientific laws and governing equations directly from data, revolutionizing the way we unravel system dynamics in numerous domains, including medicine and pharmacology. I will highlight how AI-driven methods uncover underlying principles, from classical physics to biological systems to medicine, and offer insights into future possibilities—transforming data-driven observations into interpretable and actionable scientific knowledge. Yet, can we push this boundary further—going beyond equations entirely? I will introduce direct semantic modeling, a novel paradigm where AI learns the behavior of dynamical systems directly from data without relying on closed-form equations. This semantic approach offers intuitive, human-interpretable insights into system evolution, marking a transformative leap in scientific discovery. (This talk is based on recent research with Krzysztof Kacprzyk, Tennison Liu and Sam Holt.)